# Agronomy Journal - Article

1. Vol. 101 No. 6, p. 1511-1521
OPEN ACCESS

Received: May 13, 2008

* Corresponding author(s):
View
Permissions
Share

doi:10.2134/agronj2008.0163

# Light and Moisture Competition Effects on Biomass of Red Clover Underseeded To Winter Wheat

2. Hugh Earl and
3. William Deen *
1. Dep. of Plant Agriculture, Crop Science Building, Univ. of Guelph, Guelph, ON, Canada, N1G 2W1

## Abstract

Red clover (Trifolium pratense L.) use as an underseeded cover crop in winter cereals has declined due to inability of growers to consistently establish uniform stands. The objective of this study was to assess the effect of light and soil moisture competition on underseeded red clover establishment and end of season dry matter production. Field trials were conducted at multiple locations in 2005 and 2006 in Ontario, Canada. Wheat (Triticum aestivum L.) N rate (67 and 135 kg N ha−1) and row thinning treatments (19-cm rows, every third 19-cm row removed at the 4–5 leaf stage) were used to alter light penetration and soil moisture competition. The high N rate and row thinning treatments consistently reduced light penetration, beginning as early as wheat stem elongation initiation, but had no effect on soil gravimetric moisture content. Soil moisture was primarily affected by location and year. Red clover dry weight in 2005, a relatively dry year, ranged from 688 to 1184 kg ha−1, and in 2006, a relatively wet year, ranged from 2336 to 2805 kg ha−1 Average final red clover stand count was 23 plants m−2 in 2005 and 55 plants m−2 in 2006. In 2005, plant mortality occurred before wheat anthesis. In both years, and at most locations, red clover final dry weight was positively correlated with light penetration, again beginning as early as initiation of wheat stem elongation. Final red clover dry weight in both years and red clover stand count in 2005 were correlated with soil gravimetric water content during wheat anthesis, but this was primarily due to location and year effects. Although both light penetration through the wheat canopy and soil moisture influence biomass production of underseeded red clover, soil moisture has the greater influence and is altered very little by wheat management.

Red clover is a short-lived herbaceous perennial. In Ontario and the midwestern United States it is often underseeded into an established winter cereal stand (Stute and Posner, 1993; Hesterman et al., 1992). Following cereal harvest, red clover accumulates biomass and is then either chemically or mechanically controlled in the fall or the following spring. Red clover typically precedes corn (Zea mays L.) in a rotation, where the N credit from the clover can offset some of the corn's large fertilizer N requirement. Spring underseeding of clover into established winter cereals typically has no negative impact on the yield of the winter cereal crop (Hartl, 1989; Hesterman et al., 1992; Raimbault and Vyn, 1992; Tiffin and Hesterman, 1998; Thiessen Martens et al., 2001).

Red clover provides rotational benefits when preceding corn. Nitrogen requirements of a subsequent corn crop can be significantly reduced (Vyn et al., 2000; Janovicek and Stewart, 2004; Hesterman et al., 1992). For example, in a recent review of Ontario corn N response data, 28 site-years of data were compiled for which paired comparisons of red clover versus no red clover could be made. From this analysis it was estimated that, relative to a preceding corn crop, underseeded red clover provided a N credit of 82 kg N ha−1 to corn when underseeded into small grains and incorporated with some type of tillage (Janovicek and Stewart, 2004). Red clover is also associated with increased corn yield and economic returns independent of the N fertility effect. Meyer-Aurich et al. (2006) demonstrated that yield of corn in rotation with soybean [Glycine max (L.) Merr.] and wheat increased by 5% when wheat was underseeded with red clover, and annual net returns increased by $51 to$64 ha−1 Other documented benefits of red clover in rotation include reductions in potential leaching losses of NO3 over winter months, a benefit resulting from a combined effect of red clover on water budgets and direct NO3 immobilization into biomass (Bergstrom and Kirchmann, 2004; Serran, 2005). Finally, improvements in soil structure are often associated with red clover inclusion in a rotation. Red clover increased wet aggregate stability when undersown to barley or wheat (Raimbault and Vyn, 1992).

Despite significant N and agronomic benefits associated with red clover, its use as a cover crop established by underseeding to winter wheat has declined dramatically in recent years. This observed decline is partially due to red clover stand nonuniformity with variation across a field ranging from zero red clover stand/biomass to excellent red clover stand/biomass (personal communication, P. Johnson, 2005). High levels of spatial variation in red clover biomass across a field make it impossible to apply a spatially uniform N credit to the subsequent corn crop. Due to uncertainties and costs associated with variable N applications, growers that plant corn following nonuniform stands of red clover often apply uniform rates of N. By ignoring the N credit in areas where red clover establishment and growth was good, N is overapplied, thereby increasing potential for harmful environmental losses of N in those areas.

Spatial variation in red clover stand could be due to moisture and light competition effects with winter wheat. Precipitation and soil moisture levels were observed to influence establishment of small seeded legumes (Keeling et al., 1996) and red clover underseeded to wheat (Singer et al., 2006). Drought conditions impede establishment and biomass accumulation. Light competition, in addition to moisture competition, can reduce biomass of underseeded cover crops that are planted simultaneously with winter wheat (Carofa et al., 2007).

It has been hypothesized that increased incidence of red clover spatial nonuniformity is due to increased competitiveness of winter wheat for light and soil moisture. The objective of this study was to assess, under field conditions, the effect of light and soil moisture competition on red clover establishment and end of season dry matter production.

## MATERIALS AND METHODS

### Site Description

Six winter wheat fields across Southwestern Ontario were selected for each of the 2005 and 2006 field seasons. Soil textures of all locations were within a clay loam to sandy loam category (Table 1 ), with the majority of locations planted using a no tillage system. Winter wheat was planted in 19-cm rows at 300 to 400 seeds m−2 on the planting dates indicated in Table 1 Red clover seed was broadcast applied at a rate of 450 to 750 seeds m−2 on the dates indicated in Table 1 At all locations, red clover was broadcast onto frozen soil using a spinner spreader mounted onto an all terrain vehicle when winter wheat was just initiating regrowth in the spring.

View Full Table | Close Full ViewTable 1.

Management and sampling dates by location, 2005 and 2006.

 Winter wheat planting date Winter wheat seeding rate Tillage conducted before winter wheat Red clover sowing date Red clover seeding rate Winter wheat yield harvest Red clover sampling dates† Year Field Soil texture Week 1 Week 3 Week 5 Week 7 Week 10 Week14 No. ha−1 × 106 No. ha−1 × 106 2005 BBS silt loam 14 Oct. 2004 3.9 Coulter cart/drill 5 Apr. 6.0 25 July 26 May 8 June 23 June 15 July 25 Aug. 27 Sept. DSC loam 3 Oct. 2004 3.8 light discing 26 Mar. 7.5 21 July 19 May 5 June 21 June 6 July 24 Aug. 13 Oct. DSF silt-sandy loam 2 Oct. 2004 3.4 light discing 25 Mar. 7.5 21 July 17 May 3 June 20 June 5 July 23 Aug. 22 Sept. DVG silt-clay loam 13 Oct. 2004 4.2 no-till 6 Apr. 7.5 21 July 27 May 9 June 27 June 7 July 26 Aug. 10 Oct. PJF loam-silt loam 11 Oct. 2004 3.5 no-till 13 Apr. 4.5 22 July 20 May 7 June 24 June 12 July 24 Aug. N/A SFP silt loam 12 Oct. 2004 3.0 no-till 9 Apr. 4.5 23 July 25 May 7 June 22 June 11 July 25 Aug. 3 Oct. 2006 CTB silt loam 10 Oct. 2005 3.9 no-till 4 Apr. 5.4 18 July 15 May 31 May 15 June 29 June 18 Aug. 26 Sept. CWR clay laom 6 Oct. 2005 3.7 no-till 30 Apr. 4.5 17 July 17 May 31 May 14 June 3 July 17 Aug. 21 Sept. DHG sandy loam 7 Oct. 2005 4.0 no-till 9 Mar. 4.5 19 July 12 May 29 May 14 June 27 June 18 Aug. 3 Oct. DHY silt loam 5 Oct. 2005 4.0 no-till 10 Mar. 4.5 17 July 10 May 29 May 13 June 27 June 19 Aug. 5 Oct. DSH loam-silt loam 4 Oct. 2005 4.0 no-till 11 Mar. 6.4 24 July 16 May 30 May 12 June 28 June 21 Aug. 28 Sept. PJF silt-clay loam 7 Oct. 2005 3.7 no-till 25 Mar. 7.4 18 July 15 May 26 May 9 June 26 June 21 Aug. 23 Sept.
Red clover sampling dates correspond with the following winter wheat Zadoks growth stage; Week 1, ZGS-30; Week 3, ZGS-50; Week 5, ZGS-65; Week 7, ZGS-80.

### Experimental Design

Winter wheat N rate and thinning treatments were used to alter light penetration through the wheat canopy and the soil moisture environment of underseeded red clover. In 2005, at each location, field length N rate strips of 67 kg N ha−1 and 135 kg N ha−1, the width of a commercial N applicator, were applied in the spring following initiation of winter wheat regrowth. Within each field, four positions were established such that half of each position was within the low N rate and the other half in the high N rate. At each position, within each N rate, two replications of a nonthinned and thinned treatment were established (Fig. 1 ). For the nonthinned treatment, winter wheat row spacing was maintained at 19 cm. For the thinned treatment, every third winter wheat row was chemically removed using a back-pack sprayer with a flat fan nozzle to apply Assure II (DuPont Canada Agricultural Products) (a.i. quizalofop-p-ethyl 96 g L−1) at a rate of 0.75 L ha−1 and the surfactant Sure Mix (DuPont Canada Agricultural Products, Mississauga, ON, Canada) at 0.5% v v−1 To reduce risk of herbicide injury to adjacent wheat rows, the nozzle was positioned to apply parallel with the rows and a plastic cup was used as a nozzle shroud. Chemical thinning was performed when winter wheat was at the 4–5 leaf stage with few tillers.

Fig. 1.

2005 experimental design for a single position within a location. Two replicates of thinning treatments in nonrandomized 60 and 120 kg N ha−1 strips. X denotes a randomized sample area. See Materials and Methods section for a detailed description.

In 2006, the same N rate and thinning treatments were used, however, unlike 2005, the N rate treatments were also replicated and randomized at each field site. At each of the four positions, a high and a low N application rate were randomly applied in 3- by 9-m strips. Nitrogen treatments were imposed by applying ammonium nitrate at 67 kg N ha−1 and 135 kg N ha−1 using a tractor mounted air flow fertilizer spreader. Thinned and unthinned treatments were randomly imposed on the N application strips such that the four N rate × thinning treatment combinations were arranged as a randomized complete block, and this was repeated twice at each of the four positions within a field site.

Plots were sampled once every 2 wk at each location beginning in mid-May and continuing until the winter wheat was harvested. Treatments were also sampled immediately following wheat harvest and before killing of the red clover either by tillage or herbicide. The specific sampling dates for each field location are presented in Table 1 Sampling areas for red clover count dates, end of season red clover biomass, and wheat yield were randomized, and were 112 by 50 cm, encompassing six winter wheat rows (Fig. 1).

## Data Collection

### Percentage Light Penetration

To determine photosynthetic photon flux density at the base of the canopy, a LI-191 Line Quantum Sensor (LI-COR, Lincoln NE) was placed on the soil surface parallel to the wheat row. To determine incident photosynthetic photon flux density, a LI-190 point Quantum Sensor (LI-COR, Lincoln NE) was positioned above the wheat canopy. Readings from both instruments were simultaneously recorded using a LI-1400 Data Logger (LI-COR, Lincoln NE). In 2005 and 2006, two and three measurements were taken for each sampling area on each sampling date, respectively. For both years, an average of the measurements was used to calculate mean percentage light penetration for a sample area. Light penetration measurements were not taken following winter wheat harvest.

### Gravimetric Soil Moisture

A single soil core (0–25cm) was taken from each sample area on each sample date. Soil samples were weighed and then dried at 80°C until weights measured on consecutive dates changed by <0.5%. Gravimetric soil moisture was calculated using initial sample weight and final sample weight. For each year, gravimetric soil moisture was assessed at three of the six field locations: DSC, DVG, and PJF in 2005, and DHY, DSH and PJF in 2006.

### Red Clover Stand Counts

Red clover stand count was determined by counting red clover plants in two 25-m2 quadrats in a predetermined, randomized 112- by 50-cm sample area (Fig. 1).

### Red Clover Biomass

Red clover biomass sampling was initiated on the third sampling date in both 2005 and 2006 at three of the six field locations: DSC, DVG, and PJF in 2005, and DHY, DSH, and PJF in 2006. In 2005, on the fifth and sixth sampling dates, biomass measurements were taken at all field locations. In 2006, on the fourth, fifth, and sixth sampling dates, biomass measurements were taken at all field locations. In both years, on the sixth sample week, two sample areas were harvested and averaged for the determination of final red clover biomass. Red clover biomass samples were dried in a convection drier at 80°C, until the mass of samples measured on consecutive days was observed to change by <1%.

### Wheat Grain Yield

Wheat plants were hand harvested from a 112 × 50cm sample area (Fig. 1) and threshed. Grain samples collected from the threshing process were dried in a convection drier at 80°C, until the mass measured on consecutive days was observed to change by <1%. Final grain yield was adjusted to 15% moisture content.

### Statistical Analysis

Analysis of residuals was conducted using Proc MIXED. The distribution of errors was assessed visually using Proc PLOT of residuals against predicted values, residual against replication, and residual against treatment. A Shapiro-Wilk test of residuals was conducted using Proc UNIVARIATE to ensure that residuals were normally distributed. Data that violated assumptions were subjected to either a log or square root transformation. Outlying data points were assessed using a Studentized residual.

Red clover stand counts, red clover dry weight, wheat grain yield, percentage light penetration, and soil gravimetric water content were subjected to an initial ANOVA using Proc MIXED with SAS v. 9.13 (SAS Institute, 2006). Data were analyzed by year, as a different statistical model was employed for each year. In both years, data were pooled across positions and locations. In 2005, data were analyzed as a split block design. Each N rate regime was treated as a separate field location, and data were segregated under either the high or low N application on the winter wheat. For the ANOVA, thinning, sample week, and thinning × sample week interactions were fixed effects, and all other effects were classified as random. A repeated measures statement involving the interaction between positions, replication, thinning treatment, sample week, and location was employed. In 2006, data were analyzed statistically as a randomized complete block design with two replications within each field position, and four field positions per location. Red clover failed to establish or was killed by frost at one of four positions at both the DHG and DHY locations, so data from these positions were removed from the analysis. In the model, N rate, thinning, sample week, N rate × thinning, N rate × sample week, thinning × sample week, and N rate × thinning × sample week were fixed effects, and all other effects random. A repeated measure that involved the interaction between location, position, replication, N rate treatment, and thinning treatment was again employed.

The PDIFF option was used to make comparisons between least square means on a specific sample week. The type one error rate was set at an α level of 5% (0.05).

Correlations were conducted using Proc CORR in SAS v. 9.13. Correlations were conducted using data pooled across locations and positions. Least squares means for the N rate × thinning treatments for each sample week were used. To visualize variation associated with locations rather than treatment effects, data from each location were plotted with a different symbol.

## RESULTS AND DISCUSSION

### Treatment Effects on End of Season Red Clover Dry Weight

In 2005, at all locations under both N rates, red clover final dry weight was numerically greater when winter wheat was thinned; however, the effect was only statistically significant (P < 0.05) at two locations, BBS and SFP. Averaged across locations and positions, final end of season red clover dry weight ranged from 688 to 1184 kg ha−1, with the lowest red clover biomass observed for the 135 kg N ha−1 unthinned treatment, and the highest red clover biomass observed under the 67 kg N ha−1 thinned treatment.

In 2006, plots where the winter wheat had been thinned consistently had numerically higher final red clover dry weights, but again the effect was statistically significant at only two of the locations, DHY and PJF. The lower N rate also consistently resulted in higher final red clover dry weights, with DHG and PJF statistically higher. In 2006, end of season red clover dry weight averaged across locations ranged from 2336 to 2805 kg ha−1 Lowest red clover dry weight was observed under the 135 kg N ha−1 treatment, while the highest was observed under the 67 kg N ha−1 treatment. End of season red clover dry weight was significantly reduced by 16.7% (P < 0.05) when N rate was increased from 67 kg N ha−1 to 135 kg N ha−1 Thinned treatments increased end of season red clover dry weight by 12.7% (P = 0.0006).

Final red clover dry weights observed in 2006 were comparable with a range of 980 to 1930 kg ha−1 observed in previous Ontario studies (Vyn et al., 1999, 2000; Serran, 2005). Lower red clover biomass in 2005 could be attributed to low precipitation throughout the entire growing season. In both years, final red clover dry weight was related to red clover dry weight on Sample Week 7, immediately before wheat harvest. (r = 0.77, P < 0.05 and r = 0.84, P < 0.0001 for 2005 and 2006, respectively).

### Treatment Effects on End of Season Red Clover Stand Count

In both years and across all sites, red clover effectively germinated and emerged using the broadcast, frost seeding method. Stand counts of >20 plants m−2 were observed at all sites (Fig. 2 and 3 ). This suggests that red clover nonuniformity is typically not a function of poor germination and emergence.

Fig. 2.

Red clover stand count for each week at each location in 2005. Data points are the thinning treatment means on a given sample week under each N rate. Week 1 (17–27 May) is the first sample week of data collection. * indicates thinning treatment means that are significantly different at P < 0.05.

Fig. 3.

Red clover stand count for each week at each location in 2006. Data points are the N rate and thinning treatment means on a given sample week. Week 1 (12–17 May) is the first sample week of data collection. * indicates thinning treatment means are significantly different at P < 0.05. θ indicates N rate treatment means are significantly different at P < 0.05.

Under the relatively lower rainfall conditions of 2005, stand counts tended to increase until Weeks 4 to 6, but stand count measurements taken during the period corresponding approximately with wheat anthesis were sharply reduced (Fig. 2). At a number of sites, red clover stand counts partially recovered after anthesis as a result of continued emergence and germination of red clover. Under the relatively higher rainfall conditions of 2006, stand counts were more constant throughout the season and stand count reductions at anthesis were not observed (Fig. 3). Averaged across locations, final red clover density was greater in 2006 than in 2005, an effect attributed to higher precipitation levels in that year. Similar observations that precipitation during the spring can have a substantial impact on clover establishment and stand density were made by Singer et al. (2006), Singer and Cox (1998), and Keeling et al. (1996)

At the low rate of N, final red clover density was 23 and 55 plants m−2 in 2005 and 2006, respectively. In 2005, the high rate of N reduced red clover final density by 31% (Fig. 2), although it should again be noted that N treatments were not randomized in that year. In 2006, red clover final density was reduced only at the DSH location by approximately 30% (Fig. 3). In both years, and at all locations, the thinning treatment had no effect on red clover final density. In 2006, no interaction was observed between the thinning and N treatments.

### Treatment Effects on Light Penetration

Both the thinning treatment and the low N fertilizer treatment increased light penetration through the wheat canopy. In 2005, the location effect was significant at P < 0.10 for both N rates. The location by sample week interaction was also significant for both N rates. In 2006, the location and location × sample week effects were not significant.

In 2005, lowest light penetration was experienced on Sample Week 5 with the 135 kg N ha−1 unthinned treatment at the DSF location (Fig. 4 ). Highest light penetration was observed with the 67 kg N ha−1 thinned treatment on sample week one at the PJF location. The thinning treatment significantly increased light penetration for both N fertilizer treatments at all sites. At most sites the lowest light penetration for all treatments was observed on Sample Week 5, which approximately corresponded with the time of wheat anthesis.

Fig. 4.

Light penetration through the wheat canopy for each week at each location in 2005. Data points are the thinning treatment means on a given sample week under each N rate. Week 1 (17–27 May) is the first sample week of data collection. * indicates thinning treatment means are significantly different at P < 0.05.

Averaged across all locations and treatments, light penetration was also lowest on sample week five in 2006 (Fig. 5 ). In that year, for all of the locations except CWR, light penetration was significantly higher with the 67 kg N ha−1 fertilizer treatment than the 135 kg N ha−1 treatment on several sampling dates. The thinned treatment significantly increased light penetration on each sample week for all locations. Higher minimum and maximum light penetration values observed in 2005 relative to 2006 could be a result of differences in precipitation (Table 2 ).

View Full Table | Close Full ViewTable 2.

Monthly precipitation by location, 2005 and 2006.

 2005 locations 2006 locations BBS DSC DSF DVG PJF SFP CTB CWR DSH DHG DHY PJF mm April 76 80 80 72 60 73 52 90 n/a 74 60 64 May 50 90 90 17 43 41 57 62 88 55 56 47 June 33 49 49 52 50 23 64 46 101 54 99 59 July 78 93 93 66 50 83 126 151 122 100 142 145 August 70 101 101 55 78 43 69 66 54 72 80 56 September 101 123 123 80 96 101 89 187 163 131 158 100 Total for period 409 536 536 342 376 362 456 602 441 486 599 471
Fig. 5.

Light penetration through the wheat canopy for each week at each location in 2006. Data points are the N rate and thinning treatment means on a given sample week. Week one (12–17 May) is the first sample week of data collection. * indicates thinning treatment means are significantly different at P < 0.05; θ indicates N rate treatment means are significantly different at P < 0.05.

### Treatment Effects on Soil Gravimetric Water Content

Location and year effects on soil gravimetric moisture content were much stronger than treatment effects. Effects of N rates and row thinning on soil gravimetric water content tended to be small and inconsistent. In 2005, significant sample week × location interaction effects on soil water were observed under both N rates. These effects were not statistically significant in 2006, but results are presented by location for both years for consistency (Fig. 6 and 7 ).

Fig. 6.

Soil gravimetric water content (0–22.5 cm) [g water (g dry soil)−1] for each week at each location in 2005. Data points are the thinning treatment means on a given sample week under each N rate. Week 1 (17–27 May) is the first sample week of data collection. * indicates thinning treatment means are significantly different at P < 0.05; WH indicates approximate date of winter wheat harvest.

Fig. 7.

Soil gravimetric water content (0–22.5 cm) [g water (g dry soil)−1] for each week at each location in 2006. Data points are the N rate and thinning treatment means on a given sample week. Week 1 (12–17 May) is the first sample week of data collection. * indicates thinning treatment means are significantly different at P < 0.05; θ indicates N rate treatment means are significantly different at P < 0.05; WH indicates the approximate date of winter wheat harvest.

In 2005, it was observed that the thinning treatment could either increase or decrease the gravimetric soil water content, depending on the location and sampling date (Fig. 6). For example, on Sample Week 1 at the PJF location, thinning significantly increased soil gravimetric water content under the 67 kg ha−1 N treatment, but significantly decreased it under the 135 kg ha−1 N treatment. Also, at the DSC location under the 135 kg ha−1 N treatment, thinning the wheat stand decreased soil water content on Week 1, but increased it on Week 3. Decreases in soil gravimetric water content observed at the PJF and DSC locations with the thinning treatment are opposite of what may be expected. With decreased wheat biomass, it would be expected that increased soil gravimetric water content would be present, due to reduced wheat transpiration. However, it is possible that in those cases, the thinning treatment resulted in increased soil surface evaporation due to reduced shading of the soil by the crop canopy.

2006 soil gravimetric water content was markedly different than in 2005. In both years, early-season soil moisture was relatively high, but in 2005 the soil moisture decline up to anthesis was much more rapid and pronounced than in 2006. In 2006, treatment generally had no effect on soil gravimetric moisture before wheat harvest (Fig. 7). In contrast at the end of the season, on Sample Week 14, soil gravimetric water content was affected by both thinning and N treatments. At all three locations, the thinning treatment resulted in decreased soil water content. Differences in gravimetric water content between the thinning treatments on Sample Week 14 may be attributed to greater red clover biomass in the thinned treatment, and therefore higher soil moisture depletion by transpiration. At two of the three locations, on Sample Week 14, the 135 kg N ha−1 treatment reduced soil gravimetric water content. Although decreased soil gravimetric water content may have been expected under the 67 kg N ha−1 treatment due to greater red clover dry weight, this was not observed at the PJF and DSH locations.

### Treatment Effects on Wheat Grain Yield

Increasing the N rate applied to the winter wheat crop led to significantly greater grain dry weight (5070 versus 4370 kg ha−1, P < 0.05) across locations in 2006. Although not compared statistically, grain yields in 2005 were 14.4% greater with the 135 N application rate (5310 versus 4660 kg ha−1). Winter wheat grain yield increases with higher N applications may be attributed to increases in leaf area index, aboveground biomass, and spikes per hectare (Nielsen and Halvorsen, 1991).

The thinning treatments caused a significant reduction in grain yield across locations by 19.3% (5350 vs. 3970 kg ha−1, P < 0.05) under the 67 N treatment in 2005 and by 12.9% in 2006 (5050 vs. 4400 kg ha−1,P < 0.001). Under the 135 N treatment in 2005, a nonsignificant reduction of 13.5% was observed (5700 vs. 4930 kg ha−1).

Effect of both N and thinning treatments on grain yield appear to be due to differences in light interception since, as indicated above, treatments generally did not alter soil moisture content.

It should be noted that yield reductions caused by thinning treatments should not be compared with results from other studies that examine row spacing effects. In the present study, although great care was taken to only remove designated rows by herbicide application, the potential exists for unintended deleterious effects on yields of untreated rows. Furthermore, where row spacing is established at planting time, compensatory tillering can occur. In the present study, thinning was not conducted until the first week of May in 2005 and mid April in 2006. Therefore, it is possible that spike initiation had already been completed by this point, completely eliminating any possibility of compensatory tillering. In addition, tiller death is a function of competition within the plant for dry matter, water, and nutrients (Anderson and Garlinge, 2000). It is also possible that tiller death may have already been occurring by the time thinning was completed.

### Correlations between Red Clover Dry Weight, Red Clover Stand Count, and Light Penetration

Final red clover dry weight was positively correlated with light penetration beginning early in the season. In 2005, a positive correlation was observed on Sample Week 3, and in 2006 the correlation was significant beginning at Sample Week 1 (Fig. 8 ). For all sample weeks in both years, individual locations consistently demonstrated a similar positive correlation.

Fig. 8.

End of season (Sample Week 20) red clover dry weight (g m−2) response to mean light penetration at each sample week, 2005 and 2006. Data points are the N rate × thinned treatment interactions for each location. In 2005, the PJF location was not included as final red clover dry weight was not taken.

Early season light penetration was also correlated with red clover biomass immediately before wheat harvest. For 2005, red clover dry weight on Sample Week 7 was positively correlated with increased light penetration on Sample Weeks 1 (r = 0.60, P < 0.05) and 5 (r = 0.74, P < 0.01). In 2006, increased red clover dry weight on Sample Week 7 was associated with greater light penetration on Sample Weeks 1 (r = 0.72, P < 0.001), 3 (r = 0.51, P < 0.05), 5 (r = 0.59, P < 0.01), and 7 (r = 0.57, P = 0.01). The effect of light penetration on red clover dry weight pre-wheat-harvest may be related to initial leaf area establishment of clover plants. Photosynthesis and subsequent biomass production is largely a function of the amount of light intercepted by a crop canopy (Muchow et al., 1990).

In 2005, final red clover dry weight was consistently and positively correlated with red clover stand counts observed at Week 3 (r = 0.54, P = 0.01), 5 (r = 0.49, P = 0.03), 7 (r = 0.78, P < 0.0001), 14 (r = 0.46, P = 0.04), and 20 (r = 0.84, P < 0.0001). Red clover stand counts in 2005 were reduced under low light penetration (Fig. 2). For example, stand counts on Sample Week 7 were positively correlated with light penetration on Sample Week 7 (r = 0.42, P < 0.05) as well as Sample Week 3 (r = 0.63, P < 0.01). Such a consistent relationship was not observed in 2006. Red clover final dry weight was only correlated with red clover stand count on Week 7 (r = 0.43, P = 0.03) in 2005. For 2006, no significant correlations were found between red clover stand count and mean light penetration.

### Correlations between Red Clover Dry Weight, Red Clover Stand Count, and Soil Gravimetric Water Content

As noted above, treatment effects on soil gravimetric water content within a location were small. Locations, however, differed for soil gravimetric water content. In 2006, final red clover dry weight was correlated with soil gravimetric water content measured on Sample Week 7 (Fig. 9 ). In 2005, final red clover dry weight was extremely low at the PJF location, and when this location was removed from the analysis, red clover dry weight was again found to be correlated with gravimetric soil water content at Week 7 (r = 0.64, P < 0.09). Final red clover dry weight also tended to be positively correlated with soil gravimetric water content measured at Sample Week 5 in both 2005 (r = 0.66, P < 0.07) and 2006 (r = 0.53, P < 0.07). In all of these cases, variability between locations as opposed to variability between treatments caused the significance of the relationship.

Fig. 9.

End of season (Sample Week 20) red clover dry weight (g m−2) response to mean soil gravimetric water content at each sample week, 2005 and 2006. Data points are the N rate × thinned treatment interaction means for each location.

A similar location effect was observed for correlations between soil gravimetric water content and red clover dry weight on Sample Week 7 in 2006 (r = 0.60, P < 0.05) and in 2005 (r = 0.92, P < 0.0002), Sample Week 5 in 2005 (r = 0.87, P < 0.0002), and also Sample Week 3 in 2005 (r = 0.71, P < 0.01). Sample Week 3 represented the boot emergence stage of wheat, a period that coincides with maximum leaf area index (Frederick and Camberato, 1995). Depletion of soil moisture at the boot stage observed in this study is consistent with reports in the literature. Nielsen and Halvorson (1991) found that, as a wheat crop reaches maximum leaf area index, the evapotranspiration rate increases, leading to faster depletion of soil moisture (Giunta et al., 1995).

### CONCLUSIONS

The objective of this study was to assess, under field conditions, the effect of light and soil moisture competition on red clover establishment and end of season dry matter production when underseeded to winter wheat. Nitrogen application rate and row thinning treatments were used to alter wheat canopy light penetration as well as competition for soil moisture. Light penetration was consistently increased by lower N rates and by row thinning. Soil moisture, however, was primarily affected by location and year, and was less and inconsistently affected by N rate and row thinning.

Light penetration through the winter wheat canopy, particularly during the period approximating wheat anthesis, significantly affected end of season red clover dry matter production. Reduced light penetration caused lower red clover dry weights. The effect of reduced light penetration, however, was not as pronounced as the effect of soil moisture. Final red clover dry matter was positively correlated with soil moisture, again particularly during the period around anthesis. Unlike light penetration however, treatments imposed in this study had very small and inconsistent impacts on soil moisture. Variation in soil moisture was largely determined by location and year, which may be related to seasonal differences in precipitation, soil type, or management factors, such as tillage system or wheat cultivar.

Given the apparent role of light and moisture, various management practices could be considered to enhance final red clover dry weight. Adjusting N rates to account for red clover benefits to subsequent crops in a rotation, altered wheat row spacing, or winter wheat variety selection are some of the wheat management options to consider. Early and timely frost seeding of red clover could also be used to minimize light competition effects, and possibly increase tolerance to drought stress by enhancing red clover root development before onset of moisture stress. Early seeding of red clover however, may increase the risk of mortality caused by frost (Meyer and Badaruddin, 2001), so screening for improved cold tolerance is required.

## Footnotes

• All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher.